Abstract
A heterogeneous catalyst was developed from raw Kwale red Anthill mud by thermal treatment in a muffle furnace at 900 °C for 4 h. The resulting heterogeneous catalyst was highly porous with a surface area of 42.16m2/g, possessing excellent stability as well as high catalytic activity. Central Composite Design and Machine Learning approach (Python code) were applied to model and optimize biodiesel yield from extracted watermelon oilseed. Highest biodiesel yield of 93.41 wt% was obtained under the experimental conditions of 4min duration, 350 W microwave power, 4 wt% of catalyst, and MeOH/oil ratio of 8:1 based on Central Composite Design rotatable. The optimum value of the biodiesel yield from Machine Learning was 91.7 wt%, showing a marginal performance over the Central Composite Design rotatable value (91.6 wt%) at the optimized conditions of 3 min, 280 W, 3 wt% catalyst loading and MeOH/oil molar ratio of 6:1. The correlation of the coefficient (R2) of the model was 0.9827 for Central Composite Design rotatable while the R2 of the Machine Learning model was 1.0. Thus, python coding in terms of prediction and accuracy of biodiesel yield was superior to Central Composite Design rotatable, even though both models provide a reliable response within the region of data analyzed. The Gas Chromatography-Mass Spectroscopy of the biodiesel produced revealed the presence of both saturated and unsaturated fatty acid methyl esters. Biodiesel properties from watermelon seed oil transesterification fall within the recommended standard for biodiesel fuel. This study concluded that an effective green biowaste catalyst generated from earthen waste could enhance biodiesel production from watermelon seed oil, hence, ensuring sustainability and economic feasibility for biodiesel industries.